*/
#include "seqsummarycommand.h"
-
+#include "counttable.h"
//**********************************************************************************************************************
vector<string> SeqSummaryCommand::setParameters(){
try {
- CommandParameter pfasta("fasta", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(pfasta);
- CommandParameter pname("name", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(pname);
- CommandParameter pprocessors("processors", "Number", "", "1", "", "", "",false,false); parameters.push_back(pprocessors);
- CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
- CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
+ CommandParameter pfasta("fasta", "InputTypes", "", "", "none", "none", "none","summary",false,true,true); parameters.push_back(pfasta);
+ CommandParameter pname("name", "InputTypes", "", "", "namecount", "none", "none","",false,false,true); parameters.push_back(pname);
+ CommandParameter pcount("count", "InputTypes", "", "", "namecount", "none", "none","",false,false,true); parameters.push_back(pcount);
+ CommandParameter pprocessors("processors", "Number", "", "1", "", "", "","",false,false,true); parameters.push_back(pprocessors);
+ CommandParameter pinputdir("inputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(pinputdir);
+ CommandParameter poutputdir("outputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(poutputdir);
vector<string> myArray;
for (int i = 0; i < parameters.size(); i++) { myArray.push_back(parameters[i].name); }
try {
string helpString = "";
helpString += "The summary.seqs command reads a fastafile and summarizes the sequences.\n";
- helpString += "The summary.seqs command parameters are fasta, name and processors, fasta is required, unless you have a valid current fasta file.\n";
+ helpString += "The summary.seqs command parameters are fasta, name, count and processors, fasta is required, unless you have a valid current fasta file.\n";
helpString += "The name parameter allows you to enter a name file associated with your fasta file. \n";
+ helpString += "The count parameter allows you to enter a count file associated with your fasta file. \n";
helpString += "The summary.seqs command should be in the following format: \n";
helpString += "summary.seqs(fasta=yourFastaFile, processors=2) \n";
helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFastaFile).\n";
exit(1);
}
}
+//**********************************************************************************************************************
+string SeqSummaryCommand::getOutputPattern(string type) {
+ try {
+ string pattern = "";
+
+ if (type == "summary") { pattern = "[filename],summary"; }
+ else { m->mothurOut("[ERROR]: No definition for type " + type + " output pattern.\n"); m->control_pressed = true; }
+
+ return pattern;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "SeqSummaryCommand", "getOutputPattern");
+ exit(1);
+ }
+}
//**********************************************************************************************************************
SeqSummaryCommand::SeqSummaryCommand(){
//if the user has not given a path then, add inputdir. else leave path alone.
if (path == "") { parameters["name"] = inputDir + it->second; }
}
+
+ it = parameters.find("count");
+ //user has given a template file
+ if(it != parameters.end()){
+ path = m->hasPath(it->second);
+ //if the user has not given a path then, add inputdir. else leave path alone.
+ if (path == "") { parameters["count"] = inputDir + it->second; }
+ }
}
//initialize outputTypes
if (namefile == "not open") { namefile = ""; abort = true; }
else if (namefile == "not found") { namefile = ""; }
else { m->setNameFile(namefile); }
+
+ countfile = validParameter.validFile(parameters, "count", true);
+ if (countfile == "not open") { abort = true; countfile = ""; }
+ else if (countfile == "not found") { countfile = ""; }
+ else { m->setCountTableFile(countfile); }
+
+ if ((countfile != "") && (namefile != "")) { m->mothurOut("You must enter ONLY ONE of the following: count or name."); m->mothurOutEndLine(); abort = true; }
//if the user changes the output directory command factory will send this info to us in the output parameter
outputDir = validParameter.validFile(parameters, "outputdir", false); if (outputDir == "not found"){
string temp = validParameter.validFile(parameters, "processors", false); if (temp == "not found"){ temp = m->getProcessors(); }
m->setProcessors(temp);
- convert(temp, processors);
-
-
+ m->mothurConvert(temp, processors);
+
+ if (countfile == "") {
+ if (namefile == "") {
+ vector<string> files; files.push_back(fastafile);
+ parser.getNameFile(files);
+ }
+ }
}
}
catch(exception& e) {
//set current fasta to fastafile
m->setFastaFile(fastafile);
- string summaryFile = outputDir + m->getSimpleName(fastafile) + ".summary";
+ map<string, string> variables;
+ variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(fastafile));
+ string summaryFile = getOutputFileName("summary",variables);
int numSeqs = 0;
vector<int> longHomoPolymer;
if (namefile != "") { nameMap = m->readNames(namefile); }
+ else if (countfile != "") {
+ CountTable ct;
+ ct.readTable(countfile, false, false);
+ nameMap = ct.getNameMap();
+ }
if (m->control_pressed) { return 0; }
int tag = 2001;
int startTag = 1; int endTag = 2; int lengthTag = 3; int baseTag = 4; int lhomoTag = 5;
int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
- vector<unsigned long int> MPIPos;
+ vector<unsigned long long> MPIPos;
MPI_Status status;
MPI_Status statusOut;
MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
#else
- vector<unsigned long int> positions;
- #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+ vector<unsigned long long> positions;
+ #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
positions = m->divideFile(fastafile, processors);
for (int i = 0; i < (positions.size()-1); i++) { lines.push_back(new linePair(positions[i], positions[(i+1)])); }
#else
positions = m->setFilePosFasta(fastafile, numSeqs);
+ if (positions.size() < processors) { processors = positions.size(); }
//figure out how many sequences you have to process
int numSeqsPerProcessor = numSeqs / processors;
sort(ambigBases.begin(), ambigBases.end());
sort(longHomoPolymer.begin(), longHomoPolymer.end());
int size = startPosition.size();
+
+ //find means
+ unsigned long long meanStartPosition, meanEndPosition, meanSeqLength, meanAmbigBases, meanLongHomoPolymer;
+ meanStartPosition = 0; meanEndPosition = 0; meanSeqLength = 0; meanAmbigBases = 0; meanLongHomoPolymer = 0;
+ for (int i = 0; i < size; i++) {
+ meanStartPosition += startPosition[i];
+ meanEndPosition += endPosition[i];
+ meanSeqLength += seqLength[i];
+ meanAmbigBases += ambigBases[i];
+ meanLongHomoPolymer += longHomoPolymer[i];
+ }
+
+ double meanstartPosition, meanendPosition, meanseqLength, meanambigBases, meanlongHomoPolymer;
+
+ meanstartPosition = meanStartPosition / (double) size; meanendPosition = meanEndPosition /(double) size; meanlongHomoPolymer = meanLongHomoPolymer / (double) size; meanseqLength = meanSeqLength / (double) size; meanambigBases = meanAmbigBases /(double) size;
int ptile0_25 = int(size * 0.025);
int ptile25 = int(size * 0.250);
if (m->control_pressed) { m->mothurRemove(summaryFile); return 0; }
m->mothurOutEndLine();
- m->mothurOut("\t\tStart\tEnd\tNBases\tAmbigs\tPolymer"); m->mothurOutEndLine();
- m->mothurOut("Minimum:\t" + toString(startPosition[0]) + "\t" + toString(endPosition[0]) + "\t" + toString(seqLength[0]) + "\t" + toString(ambigBases[0]) + "\t" + toString(longHomoPolymer[0])); m->mothurOutEndLine();
- m->mothurOut("2.5%-tile:\t" + toString(startPosition[ptile0_25]) + "\t" + toString(endPosition[ptile0_25]) + "\t" + toString(seqLength[ptile0_25]) + "\t" + toString(ambigBases[ptile0_25]) + "\t"+ toString(longHomoPolymer[ptile0_25])); m->mothurOutEndLine();
- m->mothurOut("25%-tile:\t" + toString(startPosition[ptile25]) + "\t" + toString(endPosition[ptile25]) + "\t" + toString(seqLength[ptile25]) + "\t" + toString(ambigBases[ptile25]) + "\t" + toString(longHomoPolymer[ptile25])); m->mothurOutEndLine();
- m->mothurOut("Median: \t" + toString(startPosition[ptile50]) + "\t" + toString(endPosition[ptile50]) + "\t" + toString(seqLength[ptile50]) + "\t" + toString(ambigBases[ptile50]) + "\t" + toString(longHomoPolymer[ptile50])); m->mothurOutEndLine();
- m->mothurOut("75%-tile:\t" + toString(startPosition[ptile75]) + "\t" + toString(endPosition[ptile75]) + "\t" + toString(seqLength[ptile75]) + "\t" + toString(ambigBases[ptile75]) + "\t" + toString(longHomoPolymer[ptile75])); m->mothurOutEndLine();
- m->mothurOut("97.5%-tile:\t" + toString(startPosition[ptile97_5]) + "\t" + toString(endPosition[ptile97_5]) + "\t" + toString(seqLength[ptile97_5]) + "\t" + toString(ambigBases[ptile97_5]) + "\t" + toString(longHomoPolymer[ptile97_5])); m->mothurOutEndLine();
- m->mothurOut("Maximum:\t" + toString(startPosition[ptile100]) + "\t" + toString(endPosition[ptile100]) + "\t" + toString(seqLength[ptile100]) + "\t" + toString(ambigBases[ptile100]) + "\t" + toString(longHomoPolymer[ptile100])); m->mothurOutEndLine();
- if (namefile == "") { m->mothurOut("# of Seqs:\t" + toString(numSeqs)); m->mothurOutEndLine(); }
+ m->mothurOut("\t\tStart\tEnd\tNBases\tAmbigs\tPolymer\tNumSeqs"); m->mothurOutEndLine();
+ m->mothurOut("Minimum:\t" + toString(startPosition[0]) + "\t" + toString(endPosition[0]) + "\t" + toString(seqLength[0]) + "\t" + toString(ambigBases[0]) + "\t" + toString(longHomoPolymer[0]) + "\t" + toString(1)); m->mothurOutEndLine();
+ m->mothurOut("2.5%-tile:\t" + toString(startPosition[ptile0_25]) + "\t" + toString(endPosition[ptile0_25]) + "\t" + toString(seqLength[ptile0_25]) + "\t" + toString(ambigBases[ptile0_25]) + "\t"+ toString(longHomoPolymer[ptile0_25]) + "\t" + toString(ptile0_25+1)); m->mothurOutEndLine();
+ m->mothurOut("25%-tile:\t" + toString(startPosition[ptile25]) + "\t" + toString(endPosition[ptile25]) + "\t" + toString(seqLength[ptile25]) + "\t" + toString(ambigBases[ptile25]) + "\t" + toString(longHomoPolymer[ptile25]) + "\t" + toString(ptile25+1)); m->mothurOutEndLine();
+ m->mothurOut("Median: \t" + toString(startPosition[ptile50]) + "\t" + toString(endPosition[ptile50]) + "\t" + toString(seqLength[ptile50]) + "\t" + toString(ambigBases[ptile50]) + "\t" + toString(longHomoPolymer[ptile50]) + "\t" + toString(ptile50+1)); m->mothurOutEndLine();
+ m->mothurOut("75%-tile:\t" + toString(startPosition[ptile75]) + "\t" + toString(endPosition[ptile75]) + "\t" + toString(seqLength[ptile75]) + "\t" + toString(ambigBases[ptile75]) + "\t" + toString(longHomoPolymer[ptile75]) + "\t" + toString(ptile75+1)); m->mothurOutEndLine();
+ m->mothurOut("97.5%-tile:\t" + toString(startPosition[ptile97_5]) + "\t" + toString(endPosition[ptile97_5]) + "\t" + toString(seqLength[ptile97_5]) + "\t" + toString(ambigBases[ptile97_5]) + "\t" + toString(longHomoPolymer[ptile97_5]) + "\t" + toString(ptile97_5+1)); m->mothurOutEndLine();
+ m->mothurOut("Maximum:\t" + toString(startPosition[ptile100]) + "\t" + toString(endPosition[ptile100]) + "\t" + toString(seqLength[ptile100]) + "\t" + toString(ambigBases[ptile100]) + "\t" + toString(longHomoPolymer[ptile100]) + "\t" + toString(ptile100+1)); m->mothurOutEndLine();
+ m->mothurOut("Mean:\t" + toString(meanstartPosition) + "\t" + toString(meanendPosition) + "\t" + toString(meanseqLength) + "\t" + toString(meanambigBases) + "\t" + toString(meanlongHomoPolymer)); m->mothurOutEndLine();
+
+ if ((namefile == "") && (countfile == "")) { m->mothurOut("# of Seqs:\t" + toString(numSeqs)); m->mothurOutEndLine(); }
else { m->mothurOut("# of unique seqs:\t" + toString(numSeqs)); m->mothurOutEndLine(); m->mothurOut("total # of seqs:\t" + toString(startPosition.size())); m->mothurOutEndLine(); }
if (m->control_pressed) { m->mothurRemove(summaryFile); return 0; }
m->mothurOutEndLine();
- m->mothurOut("Output File Name: "); m->mothurOutEndLine();
+ m->mothurOut("Output File Names: "); m->mothurOutEndLine();
m->mothurOut(summaryFile); m->mothurOutEndLine(); outputNames.push_back(summaryFile); outputTypes["summary"].push_back(summaryFile);
m->mothurOutEndLine();
}
#endif
+ //set fasta file as new current fastafile
+ string current = "";
+ itTypes = outputTypes.find("summary");
+ if (itTypes != outputTypes.end()) {
+ if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setSummaryFile(current); }
+ }
+
return 0;
}
catch(exception& e) {
bool done = false;
int count = 0;
-
+
+
while (!done) {
if (m->control_pressed) { in.close(); outSummary.close(); return 1; }
-
+
+ if (m->debug) { m->mothurOut("[DEBUG]: count = " + toString(count) + "\n"); }
+
Sequence current(in); m->gobble(in);
-
+
if (current.getName() != "") {
+ if (m->debug) { m->mothurOut("[DEBUG]: " + current.getName() + '\t' + toString(current.getNumBases()) + "\n"); }
+
int num = 1;
- if (namefile != "") {
+ if ((namefile != "") || (countfile != "")) {
//make sure this sequence is in the namefile, else error
map<string, int>::iterator it = nameMap.find(current.getName());
- if (it == nameMap.end()) { m->mothurOut("[ERROR]: " + current.getName() + " is not in your namefile, please correct."); m->mothurOutEndLine(); m->control_pressed = true; }
+ if (it == nameMap.end()) { m->mothurOut("[ERROR]: '" + current.getName() + "' is not in your name or count file, please correct."); m->mothurOutEndLine(); m->control_pressed = true; }
else { num = it->second; }
}
ambigBases.push_back(current.getAmbigBases());
longHomoPolymer.push_back(current.getLongHomoPolymer());
}
-
count++;
outSummary << current.getName() << '\t';
outSummary << current.getStartPos() << '\t' << current.getEndPos() << '\t';
outSummary << current.getNumBases() << '\t' << current.getAmbigBases() << '\t';
outSummary << current.getLongHomoPolymer() << '\t' << num << endl;
+
+ if (m->debug) { m->mothurOut("[DEBUG]: " + current.getName() + '\t' + toString(current.getNumBases()) + "\n"); }
}
- #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
- unsigned long int pos = in.tellg();
+ #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
+ unsigned long long pos = in.tellg();
if ((pos == -1) || (pos >= filePos->end)) { break; }
#else
if (in.eof()) { break; }
#endif
-
- //report progress
- //if((count) % 100 == 0){ m->mothurOut(toString(count)); m->mothurOutEndLine(); }
}
- //report progress
- //if((count) % 100 != 0){ m->mothurOut(toString(count)); m->mothurOutEndLine(); }
-
+
in.close();
return count;
}
#ifdef USE_MPI
/**************************************************************************************/
-int SeqSummaryCommand::MPICreateSummary(int start, int num, vector<int>& startPosition, vector<int>& endPosition, vector<int>& seqLength, vector<int>& ambigBases, vector<int>& longHomoPolymer, MPI_File& inMPI, MPI_File& outMPI, vector<unsigned long int>& MPIPos) {
+int SeqSummaryCommand::MPICreateSummary(int start, int num, vector<int>& startPosition, vector<int>& endPosition, vector<int>& seqLength, vector<int>& ambigBases, vector<int>& longHomoPolymer, MPI_File& inMPI, MPI_File& outMPI, vector<unsigned long long>& MPIPos) {
try {
int pid;
if (current.getName() != "") {
int num = 1;
- if (namefile != "") {
+ if ((namefile != "") || (countfile != "")) {
//make sure this sequence is in the namefile, else error
map<string, int>::iterator it = nameMap.find(current.getName());
- if (it == nameMap.end()) { cout << "[ERROR]: " << current.getName() << " is not in your namefile, please correct." << endl; m->control_pressed = true; }
+ if (it == nameMap.end()) { cout << "[ERROR]: " << current.getName() << " is not in your name or count file, please correct." << endl; m->control_pressed = true; }
else { num = it->second; }
}
//for each sequence this sequence represents
- for (int i = 0; i < num; i++) {
+ for (int j = 0; j < num; j++) {
startPosition.push_back(current.getStartPos());
endPosition.push_back(current.getEndPos());
seqLength.push_back(current.getNumBases());
int num = 0;
processIDS.clear();
-#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
//loop through and create all the processes you want
while (process != processors) {
//////////////////////////////////////////////////////////////////////////////////////////////////////
vector<seqSumData*> pDataArray;
- DWORD dwThreadIdArray[processors];
- HANDLE hThreadArray[processors];
-
+ DWORD dwThreadIdArray[processors-1];
+ HANDLE hThreadArray[processors-1];
+
+ bool hasNameMap = false;
+ if ((namefile !="") || (countfile != "")) { hasNameMap = true; }
+
//Create processor worker threads.
- for( int i=0; i<processors; i++ ){
-
- //cout << i << '\t' << lines[i]->start << '\t' << lines[i]->end << endl;
+ for( int i=0; i<processors-1; i++ ){
+
+ string extension = "";
+ if (i != 0) { extension = toString(i) + ".temp"; processIDS.push_back(i); }
// Allocate memory for thread data.
- seqSumData* tempSum = new seqSumData(&startPosition, &endPosition, &seqLength, &ambigBases, &longHomoPolymer, filename, (sumFile + toString(i) + ".temp"), m, lines[i]->start, lines[i]->end, namefile, nameMap);
+ seqSumData* tempSum = new seqSumData(filename, (sumFile+extension), m, lines[i]->start, lines[i]->end, hasNameMap, nameMap);
pDataArray.push_back(tempSum);
- processIDS.push_back(i);
-
+
//MySeqSumThreadFunction is in header. It must be global or static to work with the threads.
//default security attributes, thread function name, argument to thread function, use default creation flags, returns the thread identifier
hThreadArray[i] = CreateThread(NULL, 0, MySeqSumThreadFunction, pDataArray[i], 0, &dwThreadIdArray[i]);
}
-
+
+ //do your part
+ num = driverCreateSummary(startPosition, endPosition, seqLength, ambigBases, longHomoPolymer, fastafile, (sumFile+toString(processors-1)+".temp"), lines[processors-1]);
+ processIDS.push_back(processors-1);
+
//Wait until all threads have terminated.
- WaitForMultipleObjects(processors, hThreadArray, TRUE, INFINITE);
+ WaitForMultipleObjects(processors-1, hThreadArray, TRUE, INFINITE);
//Close all thread handles and free memory allocations.
for(int i=0; i < pDataArray.size(); i++){
num += pDataArray[i]->count;
+ if (pDataArray[i]->count != pDataArray[i]->end) {
+ m->mothurOut("[ERROR]: process " + toString(i) + " only processed " + toString(pDataArray[i]->count) + " of " + toString(pDataArray[i]->end) + " sequences assigned to it, quitting. \n"); m->control_pressed = true;
+ }
+ for (int k = 0; k < pDataArray[i]->startPosition.size(); k++) { startPosition.push_back(pDataArray[i]->startPosition[k]); }
+ for (int k = 0; k < pDataArray[i]->endPosition.size(); k++) { endPosition.push_back(pDataArray[i]->endPosition[k]); }
+ for (int k = 0; k < pDataArray[i]->seqLength.size(); k++) { seqLength.push_back(pDataArray[i]->seqLength[k]); }
+ for (int k = 0; k < pDataArray[i]->ambigBases.size(); k++) { ambigBases.push_back(pDataArray[i]->ambigBases[k]); }
+ for (int k = 0; k < pDataArray[i]->longHomoPolymer.size(); k++) { longHomoPolymer.push_back(pDataArray[i]->longHomoPolymer[k]); }
CloseHandle(hThreadArray[i]);
delete pDataArray[i];
}
-
- //rename((sumFile + toString(processIDS[0]) + ".temp").c_str(), sumFile.c_str());
+
//append files
for(int i=0;i<processIDS.size();i++){
m->appendFiles((sumFile + toString(processIDS[i]) + ".temp"), sumFile);
projects / mothur.git / blobdiff